Abstract
Pushing contents to users with device-to-device (D2D) data sharing is considered as a promising solution to overcome backhaul congestion. However, this model is associated with critical issues which are user selfishness in terms of sharing personal resources and security concern when connecting with strange devices. Therefore, we propose in this work the notion of random D2D connection where the probabilities that devices are connected for receiving or sharing data are manipulated by users themselves. Moreover, these probability values can also be treated as variables to be optimized. Based on that, we address the backhaul congestion issue and formulate it as a non-convex optimization problem. Two solving schemes are proposed based on primal decomposition and alternating direction method of multipliers algorithm, respectively. In addition, these methods are designed in both centralized and distributed manners. Besides that, the congestion probability expression are derived in the special case giving an upper bound for the performance of our presented solution. Numerical results are provided to illustrate the effectiveness of the proposed method.

This publication has 40 references indexed in Scilit: